Diagnostic efficiency of the NNET model in diagnosing TB in the discovery cohort
| Parameter | Training set | Test set |
| AUC | 0.91 (0.89 to 0.94) | 0.90 (0.86 to 0.94) |
| Accuracy | 0.85 (0.82 to 0.88) | 0.84 (0.79 to 0.89) |
| Sensitivity | 0.83 (0.77 to 0.87) | 0.70 (0.58 to 0.79) |
| Specificity | 0.86 (0.83 to 0.89) | 0.92 (0.86 to 0.96) |
| PPV | 0.77 (0.71 to 0.82) | 0.83 (0.72 to 0.91) |
| NPV | 0.90 (0.87 to 0.93) | 0.84 (0.78 to 0.90) |
| PLR | 5.97 (4.68 to 7.62) | 8.92 (4.97 to 16.03) |
| NLR | 0.20 (0.15 to 0.27) | 0.33 (0.24 to 0.46) |
Data in parentheses represent 95% CI.
AUC, area under the ROC curve; NLR, negative likelihood ratio; NNET, neural network; NPV, negative predictive value; PLR, positive likelihood ratio; PPV, positive predictive value; TB, tuberculosis.